April 2014
Volume 55, Issue 13
Free
ARVO Annual Meeting Abstract  |   April 2014
Relationship between Intensity Based Image Quality Indices of Retinal SD-OCT and Performance of Automated Retinal Layer Segmentation
Author Affiliations & Notes
  • Kyungmoo Lee
    Electrical and Computer Engineering, University of Iowa, Iowa City, IA
  • Gabrielle HS Buitendijk
    Ophthalmology, Erasmus Medical Center, Rotterdam, Netherlands
    Epidemiology, Erasmus Medical Center, Rotterdam, Netherlands
  • Henriet Springelkamp
    Ophthalmology, Erasmus Medical Center, Rotterdam, Netherlands
    Epidemiology, Erasmus Medical Center, Rotterdam, Netherlands
  • Milan Sonka
    Electrical and Computer Engineering, University of Iowa, Iowa City, IA
    Ophthalmology and Visual Sciences, University of Iowa, Iowa City, IA
  • Johannes R Vingerling
    Ophthalmology, Erasmus Medical Center, Rotterdam, Netherlands
    Epidemiology, Erasmus Medical Center, Rotterdam, Netherlands
  • Caroline C W Klaver
    Ophthalmology, Erasmus Medical Center, Rotterdam, Netherlands
    Epidemiology, Erasmus Medical Center, Rotterdam, Netherlands
  • Michael David Abramoff
    Ophthalmology and Visual Sciences, University of Iowa, Iowa City, IA
    Veterans Affairs Medical Center, Iowa City, IA
  • Footnotes
    Commercial Relationships Kyungmoo Lee, None; Gabrielle Buitendijk, None; Henriet Springelkamp, None; Milan Sonka, University of Iowa (P); Johannes Vingerling, None; Caroline Klaver, None; Michael Abramoff, IDx (E), IDx (I), University of Iowa (P)
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science April 2014, Vol.55, 4789. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Kyungmoo Lee, Gabrielle HS Buitendijk, Henriet Springelkamp, Milan Sonka, Johannes R Vingerling, Caroline C W Klaver, Michael David Abramoff; Relationship between Intensity Based Image Quality Indices of Retinal SD-OCT and Performance of Automated Retinal Layer Segmentation. Invest. Ophthalmol. Vis. Sci. 2014;55(13):4789.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract
 
Purpose
 

To investigate the relationship between image quality and automated retinal layer segmentation performance of spectral domain optical coherence tomography (SD-OCT).

 
Methods
 

3808 macular and 2978 optic nerve head (ONH) SD-OCT (3D OCT-1000 , Topcon Europe, Netherlands) volumes (512 × 128 × 650 voxels, 6.0 × 6.0 × 2.3 mm3) were obtained from both eyes of 1128 subjects (74.7 ± 8.3 years, 41% male) randomly selected from the population-based Rotterdam study. The left eye OCT volumes were flipped in a standard fashion for analysis. The OCT image quality was quantified by calculating the histogram-based maximum tissue contrast index (mTCI) presented in the literature for each volume. Total retinal thickness for each OCT volume was automatically segmented using the Iowa Reference Algorithms (available in the public domain from http://www.biomed-imaging.uiowa.edu/downloads), and the retinal layer segmentation quality was quantified as QAt, the total number of A-scans having a total retinal thickness outside t {1,2,3} standard deviations (SDs) from the population mean for that A-scan. For the ONH volumes, the central 1.5 mm was excluded from these calculations.

 
Results
 

QAt did not show any relationship to mTCI for t = 1, 2, or 3, as shown in the scatter plots.

 
Conclusions
 

The maximum tissue contrast index (mTCI) measure of OCT quality does not predict the performance of segmentation algorithms. Segmentation performance measures using the deviation from population based atlases may be necessary to objectively quantify the validity of automated layer segmentations.

   
Keywords: 549 image processing • 550 imaging/image analysis: clinical • 551 imaging/image analysis: non-clinical  
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×